[padb-devel] [padb commit] r49 - Add a new "modes" page with examples and hence
codesite-noreply at google.com
codesite-noreply at google.com
Mon Jun 15 19:12:44 BST 2009
Author: apittman
Date: Mon Jun 15 10:54:13 2009
New Revision: 49
Added:
trunk/doc/modes.html
Modified:
trunk/doc/header.html
trunk/doc/upload_website
trunk/doc/usage.html
Log:
Add a new "modes" page with examples and hence
remove some of the text from usage.
Modified: trunk/doc/header.html
==============================================================================
--- trunk/doc/header.html (original)
+++ trunk/doc/header.html Mon Jun 15 10:54:13 2009
@@ -2,11 +2,16 @@
<body>
<center>
<h1>Parallel Application Debugger</h1>
-<a href="http://padb.pittman.org.uk/" title="Main page">Padb</a>
-<a href="usage.html" title="Command line options">usage</a>
-<a href="download.html" title="Download page">download</a>
-<a href="extensions.html" title="Proposed MPI extensions">patches</a>
-<a href="email.html" title="Contact information">mailing lists</a>
+<table width=70%>
+<tr>
+<td><a href="http://padb.pittman.org.uk/" title="Main page">Padb</a></td>
+<td><a href="/usage" title="Command line options">usage</a></td>
+<td><a href="/modes" title="Modes of operation">modes</a></td>
+<td><a href="/download" title="Download page">download</a></td>
+<td><a href="/extensions" title="Proposed MPI extensions">patches</a></td>
+<td><a href="/email" title="Contact information">mailing lists</a></td>
+</tr>
+</table>
</center>
<hr>
Added: trunk/doc/modes.html
==============================================================================
--- (empty file)
+++ trunk/doc/modes.html Mon Jun 15 10:54:13 2009
@@ -0,0 +1,877 @@
+<h1>Modes of operation.</h1>
+
+<i>Padb</i> has a number of modes of operation depending on what data you
want it to report or what action
+you want it to perform, some like <b>--kill</b> generate no output but
others can generate significant
+amounts of data. This page attempts to give a summary of available modes.
+
+<p>
+
+Some modes of operation, for example stack traces, generate information per
+process in a parallel job, some modes collate information from multiple
processes
+into a single status report for the job. For modes where information is
+process specific a number of options are offered to reduce the amount of
+information displayed to the screen. Without any of these options
specified
+<i>padb</i> will prefix each line of output with the vpid followed by a
colon ": ".
+The <b>--compress-long</b> option will print a header for each process and
+display the information for that rank below the header without any
per-line prefix.
+The <b>--compress</b> option will do the same however will also attempt to
merge
+output where multiple processes in the job report identical output into a
single
+report.
+Finally there is a <b>--tree</b> option which works well with stack traces.
+
+<p>
+
+For modes where each rank is treated independently the <b>--rank</b> can
be given to target a specific processes only,
+this option can be specified multiple times to specify multiple ranks.
+
+<p>
+
+<i>Padb</i> can be told to loop, performing the same query over and over
again. This is enabled by the <b>--watch</b> flag
+and further controlled by <b>--config-option=interval=<seconds></b>
and <b>--config-option=watch-clears-screen=<bool></b>
+options, the default values for these options are 10 and 1 respectively.
+
+<p>
+
+All examples on this page show <i>padb</i> targeting a single specific
job, either by providing a number <i>job identifier</i>
+on the command line or via the <b>--all</b> or <b>--any</b> options. See
<a href=/usage>usage</a> page for information
+on selecting which jobs.
+
+<a name=proc-summary><h2>Process state</h2></a>
+The <b>--proc-summary</b> mode shows basic information about running
+processes, presented one process per line. Users can control which
information
+is shown using the <b>--proc-format</b> option.
+<table border=1>
+<tr><td>
+<pre>
+vpid hostname pid vmsize vmrss S %cpu command
+ 7 i3 2623 160336 kB 4464 kB R 49 a.out
+ 6 i2 2616 160336 kB 4464 kB R 48 a.out
+ 5 i1 2615 160336 kB 4460 kB R 47 a.out
+ 4 fnarp 2789 160336 kB 4464 kB R 44 a.out
+ 3 i3 2622 160336 kB 4464 kB R 49 a.out
+ 2 i2 2615 160336 kB 4464 kB R 49 a.out
+ 1 i1 2614 160336 kB 4468 kB R 47 a.out
+ 0 fnarp 2788 160336 kB 4464 kB R 44 a.out
+</pre>
+</tr></td>
+</table>
+
+The config option <b>proc-sort-key</b> controls which column the table is
sorted by, the default is vpid.
+
+<a name=proc-info><h3>Per-process process state</h3></a>
+
+The <b>--proc-info</b> mode reports a much more complete report about the
state of the process but doesn't easily reduce
+when run across multiple ranks. It can be controlled by the
<b>proc-shows-proc</b> (default: 1), <b>proc-shows-task</b>
+(default: 0), <b>proc-shows-fds</b> (default:1 ) and
<b>proc-shows-maps</b> (default: 0) configuration options. This output
+also shows possible columns when running in "Process state" mode.
+<table border=1>
+<tr><td>
+<pre>
+hostname:fnarp
+exe:/home/ashley/IMB/imb/src/IMB-MPI1
+Name: IMB-MPI1
+State: R (running)
+Tgid: 7743
+Pid: 7743
+PPid: 7739
+TracerPid: 0
+Uid: 1000 1000 1000 1000
+Gid: 1000 1000 1000 1000
+FDSize: 64
+Groups: 1000
+VmPeak: 251056 kB
+VmSize: 251056 kB
+VmLck: 0 kB
+VmHWM: 99820 kB
+VmRSS: 99820 kB
+VmData: 93792 kB
+VmStk: 84 kB
+VmExe: 68 kB
+VmLib: 5320 kB
+VmPTE: 408 kB
+Threads: 1
+SigQ: 0/16382
+SigPnd: 0000000000000000
+ShdPnd: 0000000000000000
+SigBlk: 0000000000000000
+SigIgn: 0000000000000000
+SigCgt: 00000001800104e0
+CapInh: 0000000000000000
+CapPrm: 0000000000000000
+CapEff: 0000000000000000
+voluntary_ctxt_switches: 1200
+nonvoluntary_ctxt_switches: 27417502
+wchan: 0
+stat: 7743 (IMB-MPI1) R 7739 7739 2505 768 7739 4202496 35270 0 6 0 44352
110655 0 0 20 0 1 0 372663 257081344 24955 18446744073709551615 4194304
4261588 140733227100464 18446744073709551615 140479664939751 0 0 0 66784 0
0 0 17 0 0 0 0 0 0
+fd0: pipe:[56964] (0 00)
+fd1: /dev/pts/5 (0 0100002)
+fd2: pipe:[56965] (0 01)
+fd3: socket:[56977] (0 04002)
+fd4: socket:[56978] (0 02)
+fd5: socket:[56983] (0 04002)
+fd6: socket:[56985] (0 04002)
+fd7: socket:[56992] (0 04002)
+fd8: socket:[57010] (0 04002)
+fd9: socket:[57012] (0 04002)
+fd10: socket:[57016] (0 04002)
+fd11: socket:[57017] (0 04002)
+fd12: socket:[57022] (0 04002)
+fd13: socket:[57023] (0 04002)
+fd14: socket:[57024] (0 04002)
+fd15: socket:[57026] (0 04002)
+fd29: pipe:[56966] (0 01)
+pcpu: 47
+</pre>
+</tr></td>
+</table>
+
+
+<a name=stack-trace><h2>Stack traces</h2></a>
+The <b>--stack-trace</b> option, best used as shown here with the
<b>--tree</b> option
+shows stack traces for each process in the job. Stack traces are
shown "backwards" with main()
+at the top to facilitate the tree view shown here.
+<table border=1>
+<tr><td>
+<pre>
+-----------------
+[0-5] (6 processes)
+-----------------
+main() at bench/tping.c:2306
+ -----------------
+ [0-1,4] (3 processes)
+ -----------------
+ timed_ping() at bench/tping.c:345
+ elan_tportRxWait() at elan4/tportRx.c:1194
+ _elan_tportRxWait() at elan4/tportRx.c:1053
+ elan_pollWord() at common/progress.c:144
+ -----------------
+ [2-3,5] (3 processes)
+ -----------------
+ timed_ping() at bench/tping.c:340
+ elan_tportTxWait() at elan4/tportTx.c:940
+ elan_wait() at common/events.c:288
+ _elan_tportTxWait() at elan4/tportTx.c:920
+ elan_waitWord() at common/progress.c:819
+-----------------
+6 (1 processes)
+-----------------
+main() at bench/tping.c:2362
+ elan_gsync() at elan4/groupGsync.c:149
+ elan_gsyncShm() at elan4/groupGsync.c:464
+ elan_gsyncNet() at elan4/groupGsync.c:99
+ elan_gsyncGeneric() at elan4/groupGsync.c:76
+ _elan_aeventWait() at elan4/devent.c:361
+ elan_deviceCheck() at common/progress.c:847
+</pre>
+</tr></td>
+</table>
+
+The config options <b>stack-shows-locals</b> and <b>stack-shows-params</b>
can be enabled to display more information in the stack
+trace, these are disabled by default and make the tree based reporting
very difficult so are best used in conjunction with the
+<b>--rank</b> option.
+
+<p>
+Stack traces on Linux often show functions below main(), these are
automatically stripped unless the flag
+<b>--nostrip-below-main</b> is provided. Likewise <i>padb</i> knows the
core "progression" functions for several parallel
+stacks and will strip functions of the other end of stacks unless the
<b>--nostrip-above-wait</b> flag is given. The list of function names to
strip beyond can be set with the <b>stack-strip-above</b> and
<b>stack-strip-below</b> configuration options, each one
+taking a comma separated list of function names.
+
+<h2>MPI Specific modes</h3>
+
+<a name=message-queue><h3>MPI message queues</h3></a>
+The option <b>--message-queue</b> will tell <i>padb</i> to read the MPI
message
+queues from your application if possible. Here shown with the
<b>--compress</b> option.
+<table border=1>
+<tr><td>
+<pre>
+----------------
+0
+----------------
+comm0: name: 'MPI_COMM_WORLD'
+comm0: rank: '0'
+comm0: size: '2'
+comm0: id: '(nil)'
+comm0: Rank: local 0 global 0
+comm0: Rank: local 1 global 1
+comm1: name: 'MPI_COMM_SELF'
+comm1: rank: '0'
+comm1: size: '1'
+comm1: id: '0x1'
+comm2: name: 'MPI_COMM_NULL'
+comm2: rank: '-2'
+comm2: size: '0'
+comm2: id: '0x2'
+comm3: name: 'MPI COMMUNICATOR 4 DUP FROM 0'
+comm3: rank: '0'
+comm3: size: '2'
+comm3: id: '0x4'
+comm3: Rank: local 0 global 0
+comm3: Rank: local 1 global 1
+comm4: name: 'MPI COMMUNICATOR 5 DUP FROM 0'
+comm4: rank: '0'
+comm4: size: '2'
+comm4: id: '0x5'
+comm4: Rank: local 0 global 0
+comm4: Rank: local 1 global 1
+comm5: name: 'MPI COMMUNICATOR 28 SPLIT FROM 4'
+comm5: rank: '0'
+comm5: size: '1'
+comm5: id: '0x1c'
+----------------
+1
+----------------
+comm0: name: 'MPI_COMM_WORLD'
+comm0: rank: '1'
+comm0: size: '2'
+comm0: id: '(nil)'
+comm0: Rank: local 0 global 0
+comm0: Rank: local 1 global 1
+comm1: name: 'MPI_COMM_SELF'
+comm1: rank: '0'
+comm1: size: '1'
+comm1: id: '0x1'
+comm2: name: 'MPI_COMM_NULL'
+comm2: rank: '-2'
+comm2: size: '0'
+comm2: id: '0x2'
+comm3: name: 'MPI COMMUNICATOR 4 DUP FROM 0'
+comm3: rank: '1'
+comm3: size: '2'
+comm3: id: '0x4'
+comm3: Rank: local 0 global 0
+comm3: Rank: local 1 global 1
+comm4: name: 'MPI COMMUNICATOR 5 DUP FROM 0'
+comm4: rank: '1'
+comm4: size: '2'
+comm4: id: '0x5'
+comm4: Rank: local 0 global 0
+comm4: Rank: local 1 global 1
+comm5: name: 'MPI COMMUNICATOR 28 SPLIT FROM 4'
+comm5: rank: '0'
+comm5: size: '1'
+comm5: id: '0x1c'
+</pre>
+</tr></td>
+</table>
+
+<a name=deadlock><h3>Collective information</h3></a>
+If you are using a <a href=/extensions>patched MPI library</a> it's also
possible for <i>padb</i> to
+display the state of collective operations across your job using the
<b>--deadlock</b> mode.
+<table border=1>
+<tr><td>
+<pre>
+Information for group '0x4' (MPI COMMUNICATOR 4 DUP FROM 0)
+Group members [1-3] (size 4) are in call 2 to Barrier.
+Group member 0 (size 4) has completed call 1 to Barrier.
+Group member 0 (size 4) is not in a call to the collectives.
+Information for group '0x5' (MPI COMMUNICATOR 5 DUP FROM 0)
+Group member 0 (size 4) is in call 2 to Barrier.
+Group members [1-3] (size 4) have completed call 1 to Barrier.
+Group members [1-3] (size 4) are not in a call to the collectives.
+Total: 6 groups of which 2 are in use.
+</pre>
+</tr></td>
+</table>
+
+<h2>Quadrics specific modes</h2>
+The <b>--set-debug</b>, <b>--group</b> and <b>--statistics</b> modes are
QsNet specific. The <b>--deadlock</b> mode
+performs the same function as <b>--group</b> for MPI programs.
+
+<a name=mpi-watch><h2>Process watch</h2></a>
+The <b>--mpi-watch</b> mode in <i>padb</i> will cause padb to inspect the
parallel job and write a single
+line of output to the screen, each character representing a process in the
parallel job. Each character
+has a different value depending on the state of that process at the time
of sampling. This provides a quick
+way to see the state of the program and if individual ranks are blocked in
comms or actively working.
+
+<p>
+
+When used with the <b>--watch</b> and
<b>--config-option=watch-clears-screen=0</b> option it becomes possible
+to see progress of the application over time, this trace shows the popular
IMB benchmarking application, those
+familiar with it will immediately see the different stages of benchmark
and how each stage uses more and more processes
+over time with any unused ones blocking in MPI_Barrier.
+<table border=1>
+<tr><td>
+<pre>
+u: unexpected messages U: unexpected and other messages
+s: sending messages r: receiving messages m: sending and receiving
+b: Barrier B: Broadcast g: Gather G: AllGather r: reduce: R: AllReduce
+a: alltoall A: alltoalls w: waiting
+.: consuming CPU cycles ,: using CPU but no queue data -: sleeping *: error
+rrbbbbbb
+rrbbbbbb
+rrbbbbbb
+rrbbbbbb
+rrbbbbbb
+ssbbbbbb
+rsbbbbbb
+r.bbbbbb
+mmbbbbbb
+mmbbbbbb
+mmbbbbbb
+mmbbbbbb
+mmbbbbbb
+mmbbbbbb
+mmbbbbbb
+m.bbbbbb
+rrbbbbbb
+r.bbbbbb
+rrbbbbbb
+rrbbbbbb
+rrbbbbbb
+rmbbbbbb
+mmbbbbbb
+mmbbbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+mmmmbbbb
+mrmmbbbb
+mmrrbbbb
+mmrmbbbb
+rmmrbbbb
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+mmrmmmrm
+mmrmrmrm
+mmrrrmmr
+mmmmmrrr
+mmmmmrrr
+bmbmmbrb
+mmbbbbbb
+mmbbbbbb
+.mbbbbbb
+mmbbbbbb
+mmbbbbbb
+mmbbbbbb
+mmbbbbbb
+mmbbbbbb
+smbbbbbb
+mmbbbbbb
+mmbbbbbb
+mmbbbbbb
+mmbbbbbb
+mmbbbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmbbbb
+msmmbbbb
+smmsbbbb
+mmmsbbbb
+msmmbbbb
+mmmmbbbb
+mmmmbbbb
+mmmmmmmm
+mmmmmmmm
+mmmmmmmm
+mmmmmmmm
+mmmmmmmm
+mmmmmmmm
+mmmmmmmm
+mmmmmmmm
+mmmmmmmm
+mmmmmmmm
+mmmmmmmm
+mmmmmmmm
+mmmmmmmm
+mmRmmmmR
+mmmmmmmm
+mmmmmmmm
+mmmmmsss
+smmmmmms
+bRbmmssb
+mmmsmsmm
+smmmmmsm
+mmmmsmms
+bmsmmbmR
+RRbbbbbb
+RRbbbbbb
+RRbbbbbb
+RRbbbbbb
+RRbbbbbb
+RRbbbbbb
+RRbbbbbb
+RRbbbbbb
+RRbbbbbb
+RRbbbbbb
+RRRRbbbb
+RRRRbbbb
+RRRRbbbb
+RRRRbbbb
+RRRRbbbb
+RRRRbbbb
+RRRRbbbb
+RRRRbbbb
+RRRRbbbb
+RRRRbbbb
+RRRRbbbb
+RRRRbbbb
+RRRRbbbb
+RRRRbbbb
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+RRRRRRRR
+rrbbbbbb
+rrbbbbbb
+rrbbbbbb
+rrbbbbbb
+rrbbbbbb
+rrbbbbbb
+rrbbbbbb
+rrbbbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrbbbb
+rbrBbbbb
+rrrrbbbb
+rrrrbbbb
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrrrrrrr
+rrbRRrrb
+rrrrrrrr
+g.bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+.R.Rbbbb
+....bbbb
+....bbbb
+....bbbb
+....RRRR
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+..bbbbbb
+..bbbbbb
+.bbbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+bb.bbbbb
+....bbbb
+....bbbb
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+..bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+........
+........
+........
+........
+........
+....b...
+........
+.R.Rb.b.
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+........
+ggbbbbbb
+ggbbbbbb
+ggbbbbbb
+ggbbbbbb
+ggbbbbbb
+ggbbbbbb
+ggbbbbbb
+ggggbbbb
+ggggbbbb
+ggggbbbb
+ggggbbbb
+ggggbbbb
+ggggbbbb
+ggggbbbb
+ggggbbbb
+ggggbbbb
+ggggbbbb
+ggggbbbb
+ggggbbbb
+ggggbbbb
+ggggbbbb
+ggggbbbb
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+gggggggg
+.bbbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+..bbbbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+bbb.bbbb
+....bbbb
+........
+........
+........
+........
+........
+.bbbbb.b
+........
+........
+........
+.b.bb.b.
+r.b.bb.b
+..bbbbbb
+..bbbbbb
+g.bbbbbb
+.Rbbbbbb
+..bbbbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+........
+........
+........
+........
+bR.RR.Bb
+........
+........
+........
+........
+........
+........
+........
+bRbRR.Rb
+g.gbbbbb
+R.bbbbbb
+....bbbb
+RRB.bbbb
+....bbbb
+.*..bbbb
+bb.bbbbb
+RbRbbRbR
+........
+........
+........
+.*......
+R*...R..
+Rabbbbbb
+aabbbbbb
+aabbbbbb
+aabbbbbb
+aaaabbbb
+aaaabbbb
+aaaabbbb
+aaaabbbb
+aaaabbbb
+aaaabbbb
+aaaabbbb
+aaaabbbb
+aaaabbbb
+aaaabbbb
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+aaaaaaaa
+ababbaba
+aaaaa*aa
+..bbbbbb
+g.bbbbbb
+..bbbbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+....bbbb
+.*..bbbb
+....bbbb
+....bbbb
+........
+.....*..
+........
+........
+.....*..
+........
+........
+........
+........
+.b.Bb.b.
+........
+........
+........
+........
+.b.b..b.
+BBbbbbbb
+BBbbbbbb
+BBbbbbbb
+BBbbbbbb
+BBBBbbbb
+BBBBbbbb
+BBBBbbbb
+BBBBbbbb
+BBBBbbbb
+BBBBbbbb
+BBBBbbbb
+BBBBbbbb
+BBBBbbbb
+BBBBBBBB
+BBBBBBBB
+BBBBBBBB
+BBBBBBBB
+BBBBBBBB
+BBBBBBBB
+BBBBBBBB
+BBBBBBBB
+BBBBBBBB
+bbbbbbbb
+********
+</pre>
+</tr></td>
+</table>
+
Modified: trunk/doc/upload_website
==============================================================================
--- trunk/doc/upload_website (original)
+++ trunk/doc/upload_website Mon Jun 15 10:54:13 2009
@@ -7,20 +7,22 @@
# Load the password from a non-public file ;)
. ~/padb-website-password.txt
-FILES="index.html usage.html download.html email.html extensions.html"
+FILES="index usage download email extensions modes"
for FILE in $FILES
do
echo Uploading $FILE
TFILE=`mktemp`
cat header.html > $TFILE
- cat $FILE >> $TFILE
+ cat $FILE.html >> $TFILE
cat ga.html >> $TFILE
- ftp-upload --host padb.pittman.org.uk -u padb at pittman.co.uk --password
$PASSWORD --as $FILE $TFILE
+ ftp-upload --host padb.pittman.org.uk -u padb at pittman.co.uk --password
$PASSWORD --as $FILE.html $TFILE
+ ftp-upload --host padb.pittman.org.uk -u padb at pittman.co.uk --password
$PASSWORD --as $FILE/index.html $TFILE
rm $TFILE
done
-ftp-upload --host padb.pittman.org.uk -u padb at pittman.co.uk --password
$PASSWORD OpenMPI-padb-groups.patch
+ftp-upload --host padb.pittman.org.uk -u padb at pittman.co.uk --password
$PASSWORD OpenMPI-padb-groups.patch
+ftp-upload --host padb.pittman.org.uk -u padb at pittman.co.uk --password
$PASSWORD --as download/OpenMPI-padb-groups.patch OpenMPI-padb-groups.patch
echo All done.
exit 0
Modified: trunk/doc/usage.html
==============================================================================
--- trunk/doc/usage.html (original)
+++ trunk/doc/usage.html Mon Jun 15 10:54:13 2009
@@ -6,8 +6,8 @@
one resource manager installed or <i>padb</i> can't detect
the correct one use the <b>rmgr</b> <a href=#config>configuration
option</a>.
-<p>If no resource manager is found use <b>-O rmgr=local</b> and process
-identifiers will be used instead of job ids.
+<p>If no resource manager is found you can use <b>-O rmgr=local</b> and
process
+identifiers (pids) will be used instead of job ids.
<p>The <b>--list-rmgrs</b> option can be used to show a list of detected
resource managers and their active jobs.
@@ -22,12 +22,12 @@
<h3>Showing list of current jobs</h3>
To show a list of currently running jobs for a given user use the
-<b>--show-jobs</b> flag.
+<b>--show-jobs</b> option. Alternatively the <b>--list-rmgrs</b> option
+shows a list of detected resource managers along with a list of currently
active jobs for each of them.
<h3>To target all jobs</h3>
To target all jobs currently running for a given user use the <b>--all</b>
(<b>-a</b>) flag.
-
<h3>To target any jobs</h3>
To target "any" job currently running for a given user use the
<b>--any</b> (<b>-A</b>) flag.
This differs from targeting all jobs as it will exit with an error if more
@@ -44,18 +44,12 @@
collect. This option is typically used when creating bug reports to
send to third parties or to inspect a job for anomalies.
-<h2>Example</h2>
-To show a stack trace for process ID 10
-use <pre>padb -Ormgr=local --stack-trace 10</pre>
-
<hr>
-
<h1>Selecting ranks (Processes)</h1>
In modes where data for each process is reported separately it is possible
to restrict which ranks are queried, this is done via the <b>--rank</b>
option.
-Multiple ranks can be selected by specifying <b>--rank</b> multiple times
-or passing a comma separated list.
+Multiple ranks can be selected by specifying <b>--rank</b> multiple times.
<hr>
@@ -72,27 +66,9 @@
iterating over the more common padb options. For large jobs this can
generate a lot of output so redirecting to a file is recommended.
-<h2>Stack Trace</h2>
-The <b>--stack-trace</b> (<b>-x</b>) option will cause a stack trace to be
given for
-each selected rank. The stack trace generated by this call is moderately
-terse and gives one function call per line. This option is best combined
-with <b>--tree</b> (<b>-t</b>) to condense stack traces into a tree based
format.
-
-<h3>Stack Trace (Longer)</h3>
-Fuller stack traces can be obtained by setting the <a
href=#config>configuration options</a>
-stack-shows-params=1 and stack-shows-locals=1. This can help to narrow
-down a problem however does not work well with tree based stack traces
-so is recommended to be used with the <b>--rank</b> flag.
-
-<h2>Message Queues</h2>
-For MPI programs the MPI Message Queues can be shown on MPI stacks
-which support it, use the <b>--message-queue</b> (<b>-q</b>) option to
show the
-message queues.
-
-<h2>Process info</h2>
-To view information about individual processes in a job use the
-<b>--proc-info</b> option, best combined with
<b>--proc=format=<format></b>,
-for example <b>--proc-format=vp,pid,state,pcpu,vmsize,vmrss</b>
+<p>
+
+A list of avaliable modes and their descriptions can be found on the <a
href=/modes>modes</a> page.
<hr>
@@ -109,12 +85,14 @@
environment variables should be specified in upper case.
<li>Config file</li>
Config is also loaded from <b>/etc/padb.conf</b> and <b>~/.padbrc</b>.
Contact the
-<a href=email.html>mailing lists</a> for help with these files.
+<a href=/email>mailing lists</a> for help with these files.
</ul>
A list of configuration options along with their current values can be
obtained by specifying a key of help with or without a value, i.e. <b>padb
-Ohelp</b>
<h2>Example</h2>
+<table border=1>
+<tr><td>
<pre>
ashley at alpha:~$ padb --show-jobs
Error, multiple resource managers detected, use -Ormgr=<resource manager>
@@ -126,3 +104,5 @@
59315
ashley at alpha:~$
</pre>
+</td></tr>
+</table>
More information about the padb-devel
mailing list